Penalised Euclidean distance regression

A method is introduced for variable selection and prediction in linear regression problems where the number of predictors can be much larger than the number of observations. The methodology involves minimising a penalised Euclidean distance, where the penalty is the geometric mean of the $\ell_1$ an...

Full description

Bibliographic Details
Main Authors: Vasiliu, Daniel, Dey, Tanujit, Dryden, Ian L.
Format: Article
Published: Wiley 2018
Subjects:
Online Access:https://eprints.nottingham.ac.uk/48710/
_version_ 1848797829794889728
author Vasiliu, Daniel
Dey, Tanujit
Dryden, Ian L.
author_facet Vasiliu, Daniel
Dey, Tanujit
Dryden, Ian L.
author_sort Vasiliu, Daniel
building Nottingham Research Data Repository
collection Online Access
description A method is introduced for variable selection and prediction in linear regression problems where the number of predictors can be much larger than the number of observations. The methodology involves minimising a penalised Euclidean distance, where the penalty is the geometric mean of the $\ell_1$ and $\ell_2$ norms of the regression coefficients. This particular formulation exhibits a grouping effect, which is useful for model selection in high dimensional problems. Also, an important result is a model consistency theorem, which does not require an estimate of the noise standard deviation. An algorithm for estimation is described, which involves thresholding to obtain a sparse solution. Practical performances of variable selection and prediction are evaluated through simulation studies and the analysis of real datasets.
first_indexed 2025-11-14T20:10:06Z
format Article
id nottingham-48710
institution University of Nottingham Malaysia Campus
institution_category Local University
last_indexed 2025-11-14T20:10:06Z
publishDate 2018
publisher Wiley
recordtype eprints
repository_type Digital Repository
spelling nottingham-487102020-05-04T19:27:55Z https://eprints.nottingham.ac.uk/48710/ Penalised Euclidean distance regression Vasiliu, Daniel Dey, Tanujit Dryden, Ian L. A method is introduced for variable selection and prediction in linear regression problems where the number of predictors can be much larger than the number of observations. The methodology involves minimising a penalised Euclidean distance, where the penalty is the geometric mean of the $\ell_1$ and $\ell_2$ norms of the regression coefficients. This particular formulation exhibits a grouping effect, which is useful for model selection in high dimensional problems. Also, an important result is a model consistency theorem, which does not require an estimate of the noise standard deviation. An algorithm for estimation is described, which involves thresholding to obtain a sparse solution. Practical performances of variable selection and prediction are evaluated through simulation studies and the analysis of real datasets. Wiley 2018-01-22 Article PeerReviewed Vasiliu, Daniel, Dey, Tanujit and Dryden, Ian L. (2018) Penalised Euclidean distance regression. Stat, 7 (1). e175/1-e175/14. ISSN 2049-1573 Euclidean distance; grouping; penalization; prediction; regularization; sparsity; variable screening http://onlinelibrary.wiley.com/doi/10.1002/sta4.175/abstract doi:10.1002/sta4.175 doi:10.1002/sta4.175
spellingShingle Euclidean distance; grouping; penalization; prediction; regularization; sparsity; variable screening
Vasiliu, Daniel
Dey, Tanujit
Dryden, Ian L.
Penalised Euclidean distance regression
title Penalised Euclidean distance regression
title_full Penalised Euclidean distance regression
title_fullStr Penalised Euclidean distance regression
title_full_unstemmed Penalised Euclidean distance regression
title_short Penalised Euclidean distance regression
title_sort penalised euclidean distance regression
topic Euclidean distance; grouping; penalization; prediction; regularization; sparsity; variable screening
url https://eprints.nottingham.ac.uk/48710/
https://eprints.nottingham.ac.uk/48710/
https://eprints.nottingham.ac.uk/48710/